Probabilistic Assessment of Monopile Foundations Accounting for Model Uncertainty using Tobit Regression

Monopile foundations are a critical component of offshore wind turbines, representing a significant portion of project costs. Performance-based design (PBD) methodologies and risk informeddesign offer a pathway to cost reduction by estimating annual failure probabilities and fragility functions through statistical modelling of extreme environmental loads and uncertainties in soil parameters. This study addresses limitations in existing fragility function estimation methods for monopile foundations under lateral loading, introducing a framework that employs Tobit regression to handle the censored displacement data. Furthermore, it explicitly incorporates soil-structure interaction model uncertainty, including bias and variance associated with traditional p-y curve formulations. The proposed methodology is demonstrated through a case study, showing how neglecting model uncertainty leads to overestimating failure probabilities in the case of high biasfactors. By addressing these gaps, the framework enhances the reliability of PBD approaches, offering a robust tool for risk-informed, cost-effective design of offshore monopile foundations.

Keywords: Performance-based design, Monopile, Offshore wind, Fragility function, Annual failure probability.